agreed.
thats also pretty much my definition of a compilation.
yes, and i don't think it should do that non-optionally. getting called a comp has lots of ramifications in SC, that users don't necessarily want, nor want to have to do additional tagging to defeat.
i appreciate VERY MUCH your efforts in writing the code to make it an option, but they seemingly refuse to allow it.
true, but again this has SC specific ramifications not present in other apps.
and as i've pointed out, SC has a bug if the string is "Various Artists."
right, altho i haven't done that yet, so i don't know how well (or not) it works.
the problem is its hard for me to know all the ramifications of using Album Artist data, or comp tags, as far as SC is concerned, b/c often the effects are far reaching, not obvious at first, and not well documented.
the guessing reference was to the VA detection logic, not the other stuff.
but even aside from that the issue is complex, b/c maybe i want things that conflict within SC. like maybe i want something ID'd as a comp, but i don't want it segregated to a separate area. or maybe i want to sort by TPE2 but i also want it know to SC as a comp. the vagaries are endless.
i would never argue that SC should handle garbage in well. thats not the issue here.
the issue is given reasonable and common tagging does SC handle it well?
thats a good option.
Results 21 to 30 of 41
-
2008-12-10, 00:05 #21
-
2008-12-10, 00:22 #22
sorry, when its wrong half the time in terms of a meaningful result, i call it guessing, and i will continue to do so.
and moreover, the whole point is to examine the following questions:
1. should the VA logic be mandatory?
2. do better ways exist?
i would say 1. no, 2. yes.
it just boggles my mind that some people here are SO against making it optional, and using other detection methods that would be better.
mine isn't unusual.
most people use mp3s, and will have differing TPE1 values for comps, as both gracenote and freedb mark tracks that way.
if they don't have TPE2 values, and many won't, they'll get misidentified albums.
if they do have TPE2 values, they will have to enable TPE2 as album artist, and totally defeat VA detection for misidentified comps.
furthermore, they will probably have to tag TPE2 values for comps as well, to work well in other apps other than SC.
in addition, gracenote assigns TPE2 to ALL files.
these conditions aren't rare, an they all lead to problems with SC's VA detection logic. its not just a couple of crazies with "libraries like mine."
most? debatable. but certainly it should be available to those who want it, i've always said it should remain an option, but BE optional.
you harp on this a lot... this forum is FOR exactly these issues imo. i heard the same stuff (from various people) before 8001 was implemented. frankly, no one forces you to respond or read any of it. now, i very much appreciate and respect your responses and you, i am amazed by your talents, but please, the den mother act is not applicable to this forum.
-
2008-12-10, 01:01 #23Senior Member
- Join Date
- Feb 2007
- Location
- Olde London Town
- Posts
- 1,099
>when its wrong half the time in terms of a meaningful result, i call it guessing, and i will continue to do so,
Why don't you just re-implement the code to do what you want? That's what open-source is all about.
-
2008-12-10, 14:15 #24Senior Member
- Join Date
- Sep 2006
- Posts
- 610
A breakthrough of sorts. I finally get what MrSinatra doesn't understand. But let's hit some lighter notes first:
Yaay lovefest! *hugs*
Kudos to you for that. I've been in the tech business long enough to know that there's nothing vendors like to do more than blame each other for bugs. And a typical response is to open a bug with all possible vendors hoping that one of them actually fixes it. But in this case, WinAmp is the only one that can fix this. The wait may be long, but you can at least be glad you're waiting in the right line, right?
Multiple artists with the same name. Not likely but possible.
And then regarding the discussion of guessing, I have to lend support to MrSinatra. Let's say someone has a deck of cards and asks you to guess if the top card is black or red. You don't have enough information to make that determination, right? Just like SqueezeCenter when it encounters an album without COMPILATION tags. You can have a system though. You can maybe know what color the last cards were, and get some probabilities. Maybe you can see the bottom card in a reflection on your shoe. So you figure out whether red or black has a greater probability, and then you choose the color with the greatest probability. That's a system that can produce consistent and predictable results, but you're still going to guess wrong sometimes because the underlying fact is you're lacking sufficient data. And I'd call that a guess.
Imagine designing a database. That database has fields to hold certain types of data. Let's say that this database is missing a field for a fax number. Now, when the guys in data entry need to enter a phone number, they use the phone number field. But every now and then they need to enter a fax number, so they decide to use the phone number field for that too. Suddenly the data in your database isn't as reliable as it once was. You can't say with any certainty whether the data in the phone field is actually a phone number. Now you suddenly have a need for that data to be accurate (let's say it's something legal just to add pressure). Oops.
Now the PRIMARY fault lies with the database designer for not including a fax field. Even if he didn't see the need right away, he should have modified the database to include that field later.
Depending on database security, there's two things the data entry guys could have done on their own. One is to do exactly what they did, and appropriate a field to do something different than what it was designed for. The second would be to create their own custom field, use it, and tell everyone else to use it. In the first example, they violate the "phone number" namespace, and in the second example, they flaunt authority and pretend to be official database admins.
So now let's take a deep dive into extended metaphor. The ID3 database has no fax field, but hardly anyone used the phone field as designed anyway, so that became the de-facto fax field. I'd have done it differently but what's done is done and it's not that big of a deal in my opinion.
But what we're talking about (compilation, for those keeping score) is ANOTHER field that's missing: let's say a checkbox for whether it's okay for advertisers and partners to call/fax this number. The guys at Apple said "I've got it! I'll create another field called TCMP and everyone should use it!" And everyone else in the office has been sick and tired of those Apple guys telling them what to do and rolled their eyes. And the guys at WinAmp said "I've got it! If it's okay for advertisers and partners to call, we should all enter (293)582-4921 into the phone number field!"
And that's the problem, roughly. WinAmp's method obliterates potentially useful album artist information by forcing all compilations to use the same album artist. And if someone happens to have an album artist that matches the compilation value, who knows what you're supposed to do.
The important thing to note is that it's the data that matters, not the software. If the data is unambiguous, then you can always easily convert the database to match another vendor's needs. Once you start polluting the data because of some vendor's harebrained scheme, you're screwed, because now your data is no good.
And that's that. I really don't think anyone's mind is going to be changed in this thread so I'm not going to bother checking back. I've done my best.
-
2008-12-10, 15:47 #25Robin BowesGuest
VA logic discussion from elsewhere...
MrGuessing wrote:
> erland;368468 Wrote:
>> Sure, there is an obvious problem with the detection logic, but this
>> still doesn't mean it is guessing. The detection logic still does what
>> it is supposed to do and it is perfectly predictable what it will do if
>> you know the rules, this is not guessing.
>>
>> If the logic was based on a random generated number and wasn't
>> predictable it would be guessing.
>
> sorry, when its wrong half the time in terms of a meaningful result, i
> call it guessing, and i will continue to do so.
By the same logic, I shall hereafter call *you* "MrGuessing".
R.
-
2008-12-10, 15:56 #26
another mature, constructive comment.
i never said i was always right, the whole point of the forums is to vet ideas and figure things out. i'm glad for you that you never need to do so.
-
2008-12-10, 18:27 #27
VA logic discussion from elsewhere...
>And then regarding the discussion of guessing, I have to lend support
>to MrSinatra. Let's say someone has a deck of cards and asks you to
>guess if the top card is black or red. You don't have enough
>information to make that determination, right? Just like SqueezeCenter
>when it encounters an album without COMPILATION tags. You can have a
>system though.
This is a bad analogy. The scanner can see all information. If it were making decisions based on some information, then perhaps you could call it guessing. eg. if it made a decision based on only 2 out of 10 tracks on an album. But that isn't the case. It reads all information on all songs, and then decides afterwards based on sound logic whether songs are part of compilation albums.
I different analogy, involving playing cards:
A scanner can see the information on all cards in a collection. If the pattern on the back of the cards were all the same, the cards are part of the same set and belong together. If one of the cards had a different pattern on the back of the card, that card belongs to a different set. This is a compilation of cards.
Tagging songs is similar. Cards=Songs, Collection of cards=Albums, Suit=Genre. Pattern on the back of cards=album artist.
Compilations follow a set of rules, because there is no entity to map a compilation to. The information is there for those rules to work.
If someone lost a card from a pack, you could take a spare from another deck of cards, and add it into the original pack. That is a compilation of cards. If you want to play a fair game, you'd have to change the pattern on the back to match, so it would not look like a compilation of cards.
You could add post-it notes on each card to say "I am/am not part of a compilation of differing cards", but that's a pretty silly thing to do, because it is only meaningful when the collection of cards is maintained. If collections were mixed up, or a single card were taken from the pack, a post-it note saying "I am/am not part of a compilation" is not really true.
The main point is that songs, albums, album artists, genres are entities that are real things. A compilation is a relationship between collection of entities, and follows rules to determine that relationship.
-
2008-12-10, 18:38 #28
VA logic discussion from elsewhere...
>To me, this means that SqueezeCenter is NOT guessing, it has logic that
>follows a set of rules to detect if an album is a compilation or not.
>
Totally agree with your logic.
>Why do the people like all the posters in this thread spend time
>discussing this over and over again in multiple threads when we all
>know we aren't going to agree ?
>
I know, mainly illogical.
My reasoning is that I don't particularly want to see more options being added due to bugs raised that are not bugs. Extra unnecessary options that could make the software slower, more permutations to making it harder to maintain/test, etc.
At best, some of the bugs raised, which point back to forum discussions that are not clear what they are asking for, are no more than enhancement requests.
I'm hoping that some of these reported issues are noticed not as bugs but as non-issues or enhancement requests, and that they are clear such that a developer doesn't blindly implement a "solution" that affects people that are happy with the current functionality. I'm not saying that they would, if there's not a clear definition, it's likely that a bug/enhancement won't be acted upon or will be pushed back to a later release.
-
2008-12-10, 18:41 #29
so phil, answer this question...
in your opinion, SC should be designed such that EVERY artist mismatch means that something should be identified to SC as a comp?
is that your opinion?
so when i have a single guest artist on a CD, that should be identified as a comp to SC?
is that what you believe?
i mean, i'm beginning to think you think SC knows better than humans as to what is and isn't a comp.
i love how you always avoid the fact that SC gets it wrong in reality. you just stick to this myopic view about the rule it uses, and not the results the rule produces.
what good is the rule, if it produces results that aren't true? and i mean true as in the reality of the situation, not the stupid rule.
-
2008-12-10, 19:30 #30



Reply With Quote

